<p>Mesoscale models, such as the Weather Research and Forecasting (WRF) model, are now commonly used to predict wind resources, and in recent years their outputs are being used as inputs to wake models for the prediction of the production of wind farms. Also, wind farm parametrizations have been implemented in the mesoscale models but their accuracy to reproduce wind speeds and turbulent kinetic energy fields within and around wind farms is yet unknown. This is partly because they have been evaluated against wind farm power measurements directly and, generally, a lack of high-quality observations of the wind field around large wind farms. Here, we evaluate the in-built wind farm parametrization of the WRF model, the so-called Fitch scheme that works together with the MYNN2 planetary boundary layer (PBL) scheme against large-eddy simulations (LES) of wakes using a generalized actuator disk model, which was also implemented within the same WRF version. After setting both types of simulations as similar as possible so that the inflow conditions are nearly identical, preliminary results show that the velocity deficits can differ up to 50% within the same area (determined by the resolution of the mesoscale run) where the turbine is placed. In contrast, within that same area, the turbine-generated TKE is nearly identical in both simulations. We also prepare an analysis of the sensitivity of the results to the inflow wind conditions, horizontal grid resolution of both the LES and the PBL run, number of turbines within the mesoscale grid cells, surface roughness, inversion strength, and boundary-layer height.</p>
Quantifying variability of Large Eddy Simulations of very large wind farms